Aug 9, 2017 · 1 min read
Nice post! With regards to encoding the categorical data and maintaining the training status, I wrote couple of helper classes that can be integrated in a pipeline. This saves the overhead of moving around column names for example.
